**** BEGIN LOGGING AT Sat Dec 18 02:59:56 2021 Dec 18 06:52:11 JaMa: I set bitbake threads to one so it doesn't compile anything else at the same time. Dec 18 07:06:22 JaMa: I'm trying to build bleeding edge stuff (kirkstone/master) and running into a few errors. You happen to have fixes for these ready somewhere already? Dec 18 07:07:30 https://bpa.st/KPMQ Dec 18 07:56:23 OK the ofono one I managed to fix. the libpbnjson and node-gyp-native not so far Dec 18 09:48:37 Morning Dec 18 09:52:09 Herrie: do you think I'd be able to help with just building the wic image on my side? I mean, is there something visibly wrong in the result you get? (no fitImage, or wrong content, or so) Dec 18 09:52:36 I could start a build, get the result in the evening Dec 18 09:57:36 Tofe: Well I'm not sure... Trevor Woerner who maintains meta-rockchip suggested I try to use meta-rockchip and meta-arm instead of our custom bits in meta-pine64-luneos Dec 18 09:57:53 So I took bleeding edge from those (honister+, kirkstone basically) and trying to build that Dec 18 09:58:06 Which means that I'm running into some issues now Dec 18 09:58:32 is there really something vital in kirkstone for our usecase? Dec 18 09:58:53 maybe we just need a SRCREV bump here and there Dec 18 09:58:57 I could try honister but that only has arm-trusted-firmware 2.5 and Kernel 5.15. Kirkstone has kernel 5.16 so I can simply patch in the DTB for the Pinephone with 2 patches Dec 18 09:59:52 Trevor suggested to use their kernel to avoid issues in defconfig at our end from Megi's version. Which makes sense for testing a bit Dec 18 10:01:08 Megi's 5.15 kernel doesn't have the PPP DTB, but I guess I could add it Dec 18 10:04:31 Let me try to build linux-yocto with the 5.16 patches for DTB Dec 18 10:04:34 Could be they will simply work Dec 18 10:08:20 Using meta-rockchip in general means a lot cleaner bits for the PPP. It builds with the following: https://github.com/Herrie82/meta-rockchip/commit/32ba2dab88cfba8aa76ae88f3b11dd397ccad3a6 Dec 18 10:10:43 Seems megi's DTB changes for 5.16 cleanly apply to 5.15 as well Dec 18 10:10:53 So I could actually try the same for Honister, might give less issues Dec 18 10:12:17 maybe yes Dec 18 10:13:22 Not so bleeding edge, but known working kernel Dec 18 10:13:34 And I guess I can always just bump ATF to 2.6 if needed Dec 18 10:13:45 rk3399 works for other boards with yocto Dec 18 10:14:07 So simply adding the PPP DTB should be a good proof of concept to exclude some issues and trace them Dec 18 10:14:23 If I can at least get that one to boot a fitImage I can narrow it futher down Dec 18 10:25:24 Tofe: I noticed that in kirkstone one of our systemd patches doesn't apply cleanly anymore because they changed a call. Might be worth to check at some point if it works without our patch Dec 18 10:26:22 I.e. they introduced: https://github.com/Herrie82/systemd/commit/627055ce9ae66a637aa2b3eec876620aec80c705#diff-af418f265bdfb783430144ad6e457be7eaeb10e2988db0996693b8c986a2dee8 Dec 18 10:26:44 Herrie: ah, yes, that would be nice Dec 18 10:31:58 Anyway the patch was easy to adjust, but might be worth to test at some point Dec 18 10:37:24 Hmmz in Honister same problem with ATF 2.6: https://paste.ofcode.org/H6KHe3N6RUffzVuqHuGfHj Dec 18 10:37:38 Any thoughts? ATF 2.6 built for me on Hardknott Dec 18 12:44:02 5.15 kernel doesn't work very well, so will upgrade to 5.16 Dec 18 12:44:10 Lots of compile issues in DTB Dec 18 12:58:10 5.16 works in Honister easily enough though Dec 18 15:48:16 Rest seems to build, only this arm-trusted-firmware-a not in some way Dec 18 15:48:29 Will try to see if I can drop meta-arm and just do like we did previously Dec 18 15:58:18 OK that works OK at my end. Weird Dec 18 15:58:48 Talked too soon, also doesn't Dec 18 16:04:42 Ah wait seems it still picks the one from meta-arm Dec 18 16:08:11 Ah seems there were 2 recipes doing the same arm-trusted-firmware and trusted-firmware-a which caused some issues Dec 18 16:11:12 Now it builds at least Dec 18 16:11:21 So I have an image shortly and can test a bit Dec 18 16:21:58 Same reboot to BL Dec 18 16:22:00 Grr Dec 18 17:12:48 Let me try again without this all-in-one image, with just the kernel Dec 18 17:21:45 Tofe: Well that seems to work better.... Dec 18 17:21:51 Getting further too! Dec 18 17:23:23 Not booting yet, but well... Not sure what was wrong with this fitImage stuff though: https://bpa.st/XAZQ Dec 18 17:53:01 Let me now check with Megi's kernel Dec 18 18:18:41 Herrie: are you using master of meta-webos-ports? it builds fine for me with kirkstone and qemux86 Dec 18 18:21:05 e.g. wasn't systemd do_patch resolved in https://github.com/webOS-ports/meta-webos-ports/commit/accb65b982bd2446f02da9c810baa8d3d7c46b22 Dec 18 18:46:34 JaMa: Well it doesn't build because of this mainly: https://github.com/webOS-ports/meta-webos-ports/blob/master/meta-luneos/conf/layer.conf#L11 Dec 18 18:46:41 So could be you have some stuff not pushed yet? Dec 18 18:49:03 JaMa: seems I had a newer systemd in my build where another patch needed updating Dec 18 18:53:56 Herrie: honister is correct value Dec 18 18:54:20 I'm building from jansa/master branch, but it should be almost the same as master Dec 18 18:54:41 JaMa: Well it was complaining to me about it Dec 18 18:54:52 strange, I've built with latest kirkstone as in webos-ports-setup on Thu Dec 18 18:54:55 Anyway I switched back to honister for now due to build errors I couldn't solve Dec 18 18:55:39 if it was complaining about LAYERSERIES_COMPAT then you had something wrong in your layer setup Dec 18 18:55:52 Well I forked the kirkstone one... Dec 18 18:56:05 Anyway I got it to build, but had issues with the libpbnjson Dec 18 18:56:11 And couldn't solve that one Dec 18 18:56:36 Could be because I was using meta-rockchip which has some specific tune for CPU Dec 18 18:56:41 But that shouldn't be really Dec 18 18:57:46 This was my log for the compile: https://bpa.st/YBXQ Dec 18 18:59:43 The good news of today's trials and tribulations is that I got an initramfs to boot at least with honister and linux-yocto-dev 5.16+ 2 DTB patches from Megi's kernel. Now testing with Megi's kernel. The approach of pulling in meta-rockchip makes the changes a lot easier and straight forward and more tested v.s. hackidy hack solution in meta-pine64-luneos for some bits Dec 18 19:02:15 PPP doesn't but further to UI yet, but one step at a time Dec 18 19:13:33 Seems with Megi's kernel getting a lot futher :) Dec 18 19:14:37 https://bpa.st/U73Q Dec 18 19:42:51 OK rebuilding since I added some missing bits in meta-wop and meta-pine64-luneos Dec 18 20:01:40 This is with some added bits (seems the same though): https://bpa.st/IX7A Dec 18 20:01:56 I might need to tweak defconfig a bit still. It might be taking the wrong one for now Dec 18 20:09:40 Tofe: how would I go about debugging now? Dec 18 20:10:07 I seem to get a prompt but then some segfaults and cannot type anything Dec 18 20:51:30 HerrieTP: just back from my tourist day in Stockholm, let me catch up :P Dec 18 20:57:36 Tofe: Nice, I like Stockholm. I can recommend Kiev as well in case you haven't been there before. Dec 18 21:14:13 Herrie: never been to Kiev no Dec 18 21:14:29 maybe when things are a bit more calm, we'll see :) Dec 18 21:15:06 for the debug, maybe double-check console=ttyS2,1500000n8 and make sure minicom is on the same config Dec 18 21:15:51 also it would help if the initramfs was picked, as you might get usb debugging Dec 18 21:16:36 Tofe: Well as you can see in my log, it should have that Dec 18 21:16:48 I mean I get into luna-next crashing now Dec 18 21:16:55 So that means it did initramfs already Dec 18 21:17:16 Herrie: no, it booted directly off sdcard Dec 18 21:17:36 I also see "Welcome to LuneOS 1.0!" ? Dec 18 21:17:41 That's not the initramfs? Dec 18 21:17:42 Ah OK Dec 18 21:17:50 ah.... you're right ! sorry Dec 18 21:18:15 but is the Welcome output part of our initramfs ? I can't remember Dec 18 21:19:24 ah the messages are a bit shuffled Dec 18 21:20:11 Are you connecting the PPP via USB at the same time ? Dec 18 21:22:16 Well I can plug in USB Dec 18 21:23:48 Tofe: I'm debugging on a Windows machine, so no minicom. Can something like Putty work or you know an alternative? Dec 18 21:24:26 I quite don't know the available tools for this on windows :/ Dec 18 21:24:34 OK Dec 18 21:24:41 Well I could plug it into my Ubuntu machine Dec 18 21:24:42 it's possible (though that would be a little miracle) that usb ssh is already in place Dec 18 21:24:50 It's a few meters away Dec 18 21:25:32 On which port the SSH would be ? 5522? Dec 18 21:25:35 Or another? Dec 18 21:25:39 just regular 22 Dec 18 21:25:45 but via usb0 Dec 18 21:25:57 so might need to setup ip Dec 18 21:26:41 on my archlinux I need to do the following: sudo ip address add 172.16.42.1/16 dev usb0 && sudo ip link set usb0 up Dec 18 21:27:19 then I can ssh to root@172.16.42.2 Dec 18 21:28:05 OK let me try that on Ubuntu Dec 18 21:34:58 Hmmz no usb0 here it seems Dec 18 21:35:58 Any way to figure out which one it might be? Dec 18 21:36:08 lsusb shows me my serial dongle, but nothing else it seems Dec 18 21:40:05 Herrie: then it's probably not up on PPP side Dec 18 21:40:31 it could be an issue with a regulator Dec 18 21:40:31 Yeah well I got minicom at hand here too now ;) Dec 18 21:40:37 Just never used it before Dec 18 21:40:46 It could be my defconfig Dec 18 21:41:08 Herrie: minicom is like a sweet taste of the 80's computing Dec 18 21:41:14 But I guess it's using Megi's which should be pretty OK Dec 18 21:41:20 I started mid 90's ;) Dec 18 21:41:34 Somewhere DOS 6.0 and Windows 3.1 Dec 18 21:41:53 well me too, just a little bit earlier (DOS 3.3) Dec 18 21:43:20 for PP I had an issue with the ANX module not being =Y but being =M, resulting in the usb not being available with the recovery UI Dec 18 21:43:20 so if this ANX stuff is also used here, maybe double-check it Dec 18 21:44:44 see https://github.com/webOS-ports/meta-pine64-luneos/commit/7d42f47540705ff24b927180b377509d51f79176 Dec 18 21:46:14 in your logs, I see "vcc5v0_typec: disabling" near the end, which reminds me of that issue Dec 18 21:48:19 I have no ANX in my defconfig it seems Dec 18 21:51:09 Let me add that Dec 18 21:51:19 I guess I took it from before your stuff was merged Dec 18 21:52:00 it's not necessarily the same hardware Dec 18 21:52:08 but well, I don't know :) Dec 18 21:54:26 Let me take my custom defconfig at least now, though I don't recall where I got it from :P Dec 18 21:54:34 Seems it took Megi's Dec 18 21:57:45 I guess my custom one came from pmOS need to check Dec 18 22:04:13 Yeah seems pmOS with some bits and pieces added that I'm building now Dec 18 22:04:20 I guess that should be fine for us in general Dec 18 22:04:23 Let me see if it boots Dec 18 22:07:10 on my side I'll soon get to bed Dec 18 22:08:18 Same here Dec 18 22:08:23 Last trial for today Dec 18 22:08:32 Flashing SD now, should be a minute Dec 18 22:09:01 This Sandisk SDXC isn't bad Dec 18 22:09:13 Get 50MB/s via USB2 on 5 year old laptop Dec 18 22:17:21 Seems the same Dec 18 22:25:15 Log FWIW: https://bpa.st/VB6Q Dec 18 22:32:53 I now nastily removed the luna-next.service from my SDcard, so that one doesn't fail anymore, nor do subsequent ones Dec 18 22:40:35 That didn't help much except for saying that they're not crashing anymore Dec 18 22:40:54 I get the login prompt via USB serial for a bit on ttyS2 but cannot type anything there Dec 18 22:41:02 THen continues with the vcc_4g etc logging **** ENDING LOGGING AT Sun Dec 19 02:59:56 2021